| 1 | /* |
| 2 | * Builtin "git grep" |
| 3 | * |
| 4 | * Copyright (c) 2006 Junio C Hamano |
| 5 | */ |
| 6 | |
| 7 | #define USE_THE_REPOSITORY_VARIABLE |
| 8 | #define DISABLE_SIGN_COMPARE_WARNINGS |
| 9 | |
| 10 | #include "builtin.h" |
| 11 | #include "abspath.h" |
| 12 | #include "environment.h" |
| 13 | #include "gettext.h" |
| 14 | #include "hex.h" |
| 15 | #include "config.h" |
| 16 | #include "tag.h" |
| 17 | #include "tree-walk.h" |
| 18 | #include "parse-options.h" |
| 19 | #include "string-list.h" |
| 20 | #include "run-command.h" |
| 21 | #include "grep.h" |
| 22 | #include "quote.h" |
| 23 | #include "dir.h" |
| 24 | #include "pathspec.h" |
| 25 | #include "setup.h" |
| 26 | #include "submodule.h" |
| 27 | #include "submodule-config.h" |
| 28 | #include "object-name.h" |
| 29 | #include "odb.h" |
| 30 | #include "odb/source.h" |
| 31 | #include "oid-array.h" |
| 32 | #include "oidset.h" |
| 33 | #include "pager.h" |
| 34 | #include "path.h" |
| 35 | #include "promisor-remote.h" |
| 36 | #include "read-cache-ll.h" |
| 37 | #include "write-or-die.h" |
| 38 | |
| 39 | static const char *grep_prefix; |
| 40 | |
| 41 | static char const * const grep_usage[] = { |
| 42 | N_("git grep [<options>] [-e] <pattern> [<rev>...] [[--] <path>...]"), |
| 43 | NULL |
| 44 | }; |
| 45 | |
| 46 | static int recurse_submodules; |
| 47 | |
| 48 | static int num_threads; |
| 49 | |
| 50 | static pthread_t *threads; |
| 51 | |
| 52 | /* We use one producer thread and THREADS consumer |
| 53 | * threads. The producer adds struct work_items to 'todo' and the |
| 54 | * consumers pick work items from the same array. |
| 55 | */ |
| 56 | struct work_item { |
| 57 | struct grep_source source; |
| 58 | char done; |
| 59 | struct strbuf out; |
| 60 | }; |
| 61 | |
| 62 | /* In the range [todo_done, todo_start) in 'todo' we have work_items |
| 63 | * that have been or are processed by a consumer thread. We haven't |
| 64 | * written the result for these to stdout yet. |
| 65 | * |
| 66 | * The work_items in [todo_start, todo_end) are waiting to be picked |
| 67 | * up by a consumer thread. |
| 68 | * |
| 69 | * The ranges are modulo TODO_SIZE. |
| 70 | */ |
| 71 | #define TODO_SIZE 128 |
| 72 | static struct work_item todo[TODO_SIZE]; |
| 73 | static int todo_start; |
| 74 | static int todo_end; |
| 75 | static int todo_done; |
| 76 | |
| 77 | /* Has all work items been added? */ |
| 78 | static int all_work_added; |
| 79 | |
| 80 | static struct repository **repos_to_free; |
| 81 | static size_t repos_to_free_nr, repos_to_free_alloc; |
| 82 | |
| 83 | /* This lock protects all the variables above. */ |
| 84 | static pthread_mutex_t grep_mutex; |
| 85 | |
| 86 | static inline void grep_lock(void) |
| 87 | { |
| 88 | pthread_mutex_lock(&grep_mutex); |
| 89 | } |
| 90 | |
| 91 | static inline void grep_unlock(void) |
| 92 | { |
| 93 | pthread_mutex_unlock(&grep_mutex); |
| 94 | } |
| 95 | |
| 96 | /* Signalled when a new work_item is added to todo. */ |
| 97 | static pthread_cond_t cond_add; |
| 98 | |
| 99 | /* Signalled when the result from one work_item is written to |
| 100 | * stdout. |
| 101 | */ |
| 102 | static pthread_cond_t cond_write; |
| 103 | |
| 104 | /* Signalled when we are finished with everything. */ |
| 105 | static pthread_cond_t cond_result; |
| 106 | |
| 107 | static int skip_first_line; |
| 108 | |
| 109 | static void add_work(struct grep_opt *opt, struct grep_source *gs) |
| 110 | { |
| 111 | if (opt->binary != GREP_BINARY_TEXT) |
| 112 | grep_source_load_driver(gs, opt->repo->index); |
| 113 | |
| 114 | grep_lock(); |
| 115 | |
| 116 | while ((todo_end+1) % ARRAY_SIZE(todo) == todo_done) { |
| 117 | pthread_cond_wait(&cond_write, &grep_mutex); |
| 118 | } |
| 119 | |
| 120 | todo[todo_end].source = *gs; |
| 121 | todo[todo_end].done = 0; |
| 122 | strbuf_reset(&todo[todo_end].out); |
| 123 | todo_end = (todo_end + 1) % ARRAY_SIZE(todo); |
| 124 | |
| 125 | pthread_cond_signal(&cond_add); |
| 126 | grep_unlock(); |
| 127 | } |
| 128 | |
| 129 | static struct work_item *get_work(void) |
| 130 | { |
| 131 | struct work_item *ret; |
| 132 | |
| 133 | grep_lock(); |
| 134 | while (todo_start == todo_end && !all_work_added) { |
| 135 | pthread_cond_wait(&cond_add, &grep_mutex); |
| 136 | } |
| 137 | |
| 138 | if (todo_start == todo_end && all_work_added) { |
| 139 | ret = NULL; |
| 140 | } else { |
| 141 | ret = &todo[todo_start]; |
| 142 | todo_start = (todo_start + 1) % ARRAY_SIZE(todo); |
| 143 | } |
| 144 | grep_unlock(); |
| 145 | return ret; |
| 146 | } |
| 147 | |
| 148 | static void work_done(struct work_item *w) |
| 149 | { |
| 150 | int old_done; |
| 151 | |
| 152 | grep_lock(); |
| 153 | w->done = 1; |
| 154 | old_done = todo_done; |
| 155 | for(; todo[todo_done].done && todo_done != todo_start; |
| 156 | todo_done = (todo_done+1) % ARRAY_SIZE(todo)) { |
| 157 | w = &todo[todo_done]; |
| 158 | if (w->out.len) { |
| 159 | const char *p = w->out.buf; |
| 160 | size_t len = w->out.len; |
| 161 | |
| 162 | /* Skip the leading hunk mark of the first file. */ |
| 163 | if (skip_first_line) { |
| 164 | while (len) { |
| 165 | len--; |
| 166 | if (*p++ == '\n') |
| 167 | break; |
| 168 | } |
| 169 | skip_first_line = 0; |
| 170 | } |
| 171 | |
| 172 | write_or_die(1, p, len); |
| 173 | } |
| 174 | grep_source_clear(&w->source); |
| 175 | } |
| 176 | |
| 177 | if (old_done != todo_done) |
| 178 | pthread_cond_signal(&cond_write); |
| 179 | |
| 180 | if (all_work_added && todo_done == todo_end) |
| 181 | pthread_cond_signal(&cond_result); |
| 182 | |
| 183 | grep_unlock(); |
| 184 | } |
| 185 | |
| 186 | static void free_repos(void) |
| 187 | { |
| 188 | int i; |
| 189 | |
| 190 | for (i = 0; i < repos_to_free_nr; i++) { |
| 191 | repo_clear(repos_to_free[i]); |
| 192 | free(repos_to_free[i]); |
| 193 | } |
| 194 | FREE_AND_NULL(repos_to_free); |
| 195 | repos_to_free_nr = 0; |
| 196 | repos_to_free_alloc = 0; |
| 197 | } |
| 198 | |
| 199 | static void *run(void *arg) |
| 200 | { |
| 201 | int hit = 0; |
| 202 | struct grep_opt *opt = arg; |
| 203 | |
| 204 | while (1) { |
| 205 | struct work_item *w = get_work(); |
| 206 | if (!w) |
| 207 | break; |
| 208 | |
| 209 | opt->output_priv = w; |
| 210 | hit |= grep_source(opt, &w->source); |
| 211 | grep_source_clear_data(&w->source); |
| 212 | work_done(w); |
| 213 | } |
| 214 | free_grep_patterns(opt); |
| 215 | free(opt); |
| 216 | |
| 217 | return (void*) (intptr_t) hit; |
| 218 | } |
| 219 | |
| 220 | static void strbuf_out(struct grep_opt *opt, const void *buf, size_t size) |
| 221 | { |
| 222 | struct work_item *w = opt->output_priv; |
| 223 | strbuf_add(&w->out, buf, size); |
| 224 | } |
| 225 | |
| 226 | static void start_threads(struct grep_opt *opt) |
| 227 | { |
| 228 | int i; |
| 229 | |
| 230 | pthread_mutex_init(&grep_mutex, NULL); |
| 231 | pthread_mutex_init(&grep_attr_mutex, NULL); |
| 232 | pthread_cond_init(&cond_add, NULL); |
| 233 | pthread_cond_init(&cond_write, NULL); |
| 234 | pthread_cond_init(&cond_result, NULL); |
| 235 | grep_use_locks = 1; |
| 236 | enable_obj_read_lock(); |
| 237 | |
| 238 | for (i = 0; i < ARRAY_SIZE(todo); i++) { |
| 239 | strbuf_init(&todo[i].out, 0); |
| 240 | } |
| 241 | |
| 242 | CALLOC_ARRAY(threads, num_threads); |
| 243 | for (i = 0; i < num_threads; i++) { |
| 244 | int err; |
| 245 | struct grep_opt *o = grep_opt_dup(opt); |
| 246 | o->output = strbuf_out; |
| 247 | compile_grep_patterns(o); |
| 248 | err = pthread_create(&threads[i], NULL, run, o); |
| 249 | |
| 250 | if (err) |
| 251 | die(_("grep: failed to create thread: %s"), |
| 252 | strerror(err)); |
| 253 | } |
| 254 | } |
| 255 | |
| 256 | static int wait_all(void) |
| 257 | { |
| 258 | int hit = 0; |
| 259 | int i; |
| 260 | |
| 261 | if (!HAVE_THREADS) |
| 262 | BUG("Never call this function unless you have started threads"); |
| 263 | |
| 264 | grep_lock(); |
| 265 | all_work_added = 1; |
| 266 | |
| 267 | /* Wait until all work is done. */ |
| 268 | while (todo_done != todo_end) |
| 269 | pthread_cond_wait(&cond_result, &grep_mutex); |
| 270 | |
| 271 | /* Wake up all the consumer threads so they can see that there |
| 272 | * is no more work to do. |
| 273 | */ |
| 274 | pthread_cond_broadcast(&cond_add); |
| 275 | grep_unlock(); |
| 276 | |
| 277 | for (i = 0; i < num_threads; i++) { |
| 278 | void *h; |
| 279 | pthread_join(threads[i], &h); |
| 280 | hit |= (int) (intptr_t) h; |
| 281 | } |
| 282 | |
| 283 | free(threads); |
| 284 | |
| 285 | pthread_mutex_destroy(&grep_mutex); |
| 286 | pthread_mutex_destroy(&grep_attr_mutex); |
| 287 | pthread_cond_destroy(&cond_add); |
| 288 | pthread_cond_destroy(&cond_write); |
| 289 | pthread_cond_destroy(&cond_result); |
| 290 | grep_use_locks = 0; |
| 291 | disable_obj_read_lock(); |
| 292 | |
| 293 | return hit; |
| 294 | } |
| 295 | |
| 296 | static int grep_cmd_config(const char *var, const char *value, |
| 297 | const struct config_context *ctx, void *cb) |
| 298 | { |
| 299 | int st = grep_config(var, value, ctx, cb); |
| 300 | |
| 301 | if (git_color_config(var, value, cb) < 0) |
| 302 | st = -1; |
| 303 | else if (git_default_config(var, value, ctx, cb) < 0) |
| 304 | st = -1; |
| 305 | |
| 306 | if (!strcmp(var, "grep.threads")) { |
| 307 | num_threads = git_config_int(var, value, ctx->kvi); |
| 308 | if (num_threads < 0) |
| 309 | die(_("invalid number of threads specified (%d) for %s"), |
| 310 | num_threads, var); |
| 311 | else if (!HAVE_THREADS && num_threads > 1) { |
| 312 | /* |
| 313 | * TRANSLATORS: %s is the configuration |
| 314 | * variable for tweaking threads, currently |
| 315 | * grep.threads |
| 316 | */ |
| 317 | warning(_("no threads support, ignoring %s"), var); |
| 318 | num_threads = 1; |
| 319 | } |
| 320 | } |
| 321 | |
| 322 | if (!strcmp(var, "submodule.recurse")) |
| 323 | recurse_submodules = git_config_bool(var, value); |
| 324 | |
| 325 | return st; |
| 326 | } |
| 327 | |
| 328 | static void grep_source_name(struct grep_opt *opt, const char *filename, |
| 329 | int tree_name_len, struct strbuf *out) |
| 330 | { |
| 331 | strbuf_reset(out); |
| 332 | |
| 333 | if (opt->null_following_name) { |
| 334 | if (opt->relative && grep_prefix) { |
| 335 | struct strbuf rel_buf = STRBUF_INIT; |
| 336 | const char *rel_name = |
| 337 | relative_path(filename + tree_name_len, |
| 338 | grep_prefix, &rel_buf); |
| 339 | |
| 340 | if (tree_name_len) |
| 341 | strbuf_add(out, filename, tree_name_len); |
| 342 | |
| 343 | strbuf_addstr(out, rel_name); |
| 344 | strbuf_release(&rel_buf); |
| 345 | } else { |
| 346 | strbuf_addstr(out, filename); |
| 347 | } |
| 348 | return; |
| 349 | } |
| 350 | |
| 351 | if (opt->relative && grep_prefix) |
| 352 | quote_path(filename + tree_name_len, grep_prefix, out, 0); |
| 353 | else |
| 354 | quote_c_style(filename + tree_name_len, out, NULL, 0); |
| 355 | |
| 356 | if (tree_name_len) |
| 357 | strbuf_insert(out, 0, filename, tree_name_len); |
| 358 | } |
| 359 | |
| 360 | static int grep_oid(struct grep_opt *opt, const struct object_id *oid, |
| 361 | const char *filename, int tree_name_len, |
| 362 | const char *path) |
| 363 | { |
| 364 | struct strbuf pathbuf = STRBUF_INIT; |
| 365 | struct grep_source gs; |
| 366 | |
| 367 | grep_source_name(opt, filename, tree_name_len, &pathbuf); |
| 368 | grep_source_init_oid(&gs, pathbuf.buf, path, oid, opt->repo); |
| 369 | strbuf_release(&pathbuf); |
| 370 | |
| 371 | if (num_threads > 1) { |
| 372 | /* |
| 373 | * add_work() copies gs and thus assumes ownership of |
| 374 | * its fields, so do not call grep_source_clear() |
| 375 | */ |
| 376 | add_work(opt, &gs); |
| 377 | return 0; |
| 378 | } else { |
| 379 | int hit; |
| 380 | |
| 381 | hit = grep_source(opt, &gs); |
| 382 | |
| 383 | grep_source_clear(&gs); |
| 384 | return hit; |
| 385 | } |
| 386 | } |
| 387 | |
| 388 | static int grep_file(struct grep_opt *opt, const char *filename) |
| 389 | { |
| 390 | struct strbuf buf = STRBUF_INIT; |
| 391 | struct grep_source gs; |
| 392 | |
| 393 | grep_source_name(opt, filename, 0, &buf); |
| 394 | grep_source_init_file(&gs, buf.buf, filename); |
| 395 | strbuf_release(&buf); |
| 396 | |
| 397 | if (num_threads > 1) { |
| 398 | /* |
| 399 | * add_work() copies gs and thus assumes ownership of |
| 400 | * its fields, so do not call grep_source_clear() |
| 401 | */ |
| 402 | add_work(opt, &gs); |
| 403 | return 0; |
| 404 | } else { |
| 405 | int hit; |
| 406 | |
| 407 | hit = grep_source(opt, &gs); |
| 408 | |
| 409 | grep_source_clear(&gs); |
| 410 | return hit; |
| 411 | } |
| 412 | } |
| 413 | |
| 414 | static void append_path(struct grep_opt *opt, const void *data, size_t len) |
| 415 | { |
| 416 | struct string_list *path_list = opt->output_priv; |
| 417 | |
| 418 | if (len == 1 && *(const char *)data == '\0') |
| 419 | return; |
| 420 | string_list_append_nodup(path_list, xstrndup(data, len)); |
| 421 | } |
| 422 | |
| 423 | static void run_pager(struct grep_opt *opt, const char *prefix) |
| 424 | { |
| 425 | struct string_list *path_list = opt->output_priv; |
| 426 | struct child_process child = CHILD_PROCESS_INIT; |
| 427 | int i, status; |
| 428 | |
| 429 | for (i = 0; i < path_list->nr; i++) |
| 430 | strvec_push(&child.args, path_list->items[i].string); |
| 431 | child.dir = prefix; |
| 432 | child.use_shell = 1; |
| 433 | |
| 434 | status = run_command(&child); |
| 435 | if (status) |
| 436 | exit(status); |
| 437 | } |
| 438 | |
| 439 | static int grep_cache(struct grep_opt *opt, |
| 440 | const struct pathspec *pathspec, int cached); |
| 441 | static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec, |
| 442 | struct tree_desc *tree, struct strbuf *base, int tn_len, |
| 443 | int check_attr); |
| 444 | |
| 445 | static int grep_submodule(struct grep_opt *opt, |
| 446 | const struct pathspec *pathspec, |
| 447 | const struct object_id *oid, |
| 448 | const char *filename, const char *path, int cached) |
| 449 | { |
| 450 | struct repository *subrepo; |
| 451 | struct repository *superproject = opt->repo; |
| 452 | struct grep_opt subopt; |
| 453 | int hit = 0; |
| 454 | |
| 455 | if (!is_submodule_active(superproject, path)) |
| 456 | return 0; |
| 457 | |
| 458 | subrepo = xmalloc(sizeof(*subrepo)); |
| 459 | if (repo_submodule_init(subrepo, superproject, path, null_oid(opt->repo->hash_algo))) { |
| 460 | free(subrepo); |
| 461 | return 0; |
| 462 | } |
| 463 | ALLOC_GROW(repos_to_free, repos_to_free_nr + 1, repos_to_free_alloc); |
| 464 | repos_to_free[repos_to_free_nr++] = subrepo; |
| 465 | |
| 466 | /* |
| 467 | * NEEDSWORK: repo_read_gitmodules() might call |
| 468 | * odb_add_to_alternates_memory() via config_from_gitmodules(). This |
| 469 | * operation causes a race condition with concurrent object readings |
| 470 | * performed by the worker threads. That's why we need obj_read_lock() |
| 471 | * here. It should be removed once it's no longer necessary to add the |
| 472 | * subrepo's odbs to the in-memory alternates list. |
| 473 | */ |
| 474 | obj_read_lock(); |
| 475 | |
| 476 | /* |
| 477 | * NEEDSWORK: when reading a submodule, the sparsity settings in the |
| 478 | * superproject are incorrectly forgotten or misused. For example: |
| 479 | * |
| 480 | * 1. "command_requires_full_index" |
| 481 | * When this setting is turned on for `grep`, only the superproject |
| 482 | * knows it. All the submodules are read with their own configs |
| 483 | * and get prepare_repo_settings()'d. Therefore, these submodules |
| 484 | * "forget" the sparse-index feature switch. As a result, the index |
| 485 | * of these submodules are expanded unexpectedly. |
| 486 | * |
| 487 | * 2. "config_values_private_.apply_sparse_checkout" |
| 488 | * When running `grep` in the superproject, this setting is |
| 489 | * populated using the superproject's configs. However, once |
| 490 | * initialized, this config is globally accessible and is read by |
| 491 | * prepare_repo_settings() for the submodules. For instance, if a |
| 492 | * submodule is using a sparse-checkout, however, the superproject |
| 493 | * is not, the result is that the config from the superproject will |
| 494 | * dictate the behavior for the submodule, making it "forget" its |
| 495 | * sparse-checkout state. |
| 496 | * |
| 497 | * 3. "core_sparse_checkout_cone" |
| 498 | * ditto. |
| 499 | * |
| 500 | * Note that this list is not exhaustive. |
| 501 | */ |
| 502 | repo_read_gitmodules(subrepo, 0); |
| 503 | |
| 504 | /* |
| 505 | * All code paths tested by test code no longer need submodule ODBs to |
| 506 | * be added as alternates, but add it to the list just in case. |
| 507 | * Submodule ODBs added through add_submodule_odb_by_path() will be |
| 508 | * lazily registered as alternates when needed (and except in an |
| 509 | * unexpected code interaction, it won't be needed). |
| 510 | */ |
| 511 | odb_add_submodule_source_by_path(the_repository->objects, |
| 512 | subrepo->objects->sources->path); |
| 513 | obj_read_unlock(); |
| 514 | |
| 515 | memcpy(&subopt, opt, sizeof(subopt)); |
| 516 | subopt.repo = subrepo; |
| 517 | |
| 518 | if (oid) { |
| 519 | enum object_type object_type; |
| 520 | struct tree_desc tree; |
| 521 | void *data; |
| 522 | size_t size; |
| 523 | struct strbuf base = STRBUF_INIT; |
| 524 | |
| 525 | obj_read_lock(); |
| 526 | object_type = odb_read_object_info(subrepo->objects, oid, NULL); |
| 527 | obj_read_unlock(); |
| 528 | data = odb_read_object_peeled(subrepo->objects, oid, OBJ_TREE, &size, NULL); |
| 529 | if (!data) |
| 530 | die(_("unable to read tree (%s)"), oid_to_hex(oid)); |
| 531 | |
| 532 | strbuf_addstr(&base, filename); |
| 533 | strbuf_addch(&base, '/'); |
| 534 | |
| 535 | init_tree_desc(&tree, oid, data, size); |
| 536 | hit = grep_tree(&subopt, pathspec, &tree, &base, base.len, |
| 537 | object_type == OBJ_COMMIT); |
| 538 | strbuf_release(&base); |
| 539 | free(data); |
| 540 | } else { |
| 541 | hit = grep_cache(&subopt, pathspec, cached); |
| 542 | } |
| 543 | |
| 544 | return hit; |
| 545 | } |
| 546 | |
| 547 | static int grep_cache(struct grep_opt *opt, |
| 548 | const struct pathspec *pathspec, int cached) |
| 549 | { |
| 550 | struct repository *repo = opt->repo; |
| 551 | int hit = 0; |
| 552 | int nr; |
| 553 | struct strbuf name = STRBUF_INIT; |
| 554 | int name_base_len = 0; |
| 555 | if (repo->submodule_prefix) { |
| 556 | name_base_len = strlen(repo->submodule_prefix); |
| 557 | strbuf_addstr(&name, repo->submodule_prefix); |
| 558 | } |
| 559 | |
| 560 | if (repo_read_index(repo) < 0) |
| 561 | die(_("index file corrupt")); |
| 562 | |
| 563 | for (nr = 0; nr < repo->index->cache_nr; nr++) { |
| 564 | const struct cache_entry *ce = repo->index->cache[nr]; |
| 565 | |
| 566 | if (!cached && ce_skip_worktree(ce)) |
| 567 | continue; |
| 568 | |
| 569 | strbuf_setlen(&name, name_base_len); |
| 570 | strbuf_addstr(&name, ce->name); |
| 571 | if (S_ISSPARSEDIR(ce->ce_mode)) { |
| 572 | enum object_type type; |
| 573 | struct tree_desc tree; |
| 574 | void *data; |
| 575 | size_t size; |
| 576 | |
| 577 | data = odb_read_object(the_repository->objects, &ce->oid, |
| 578 | &type, &size); |
| 579 | if (!data) |
| 580 | die(_("unable to read tree %s"), oid_to_hex(&ce->oid)); |
| 581 | init_tree_desc(&tree, &ce->oid, data, size); |
| 582 | |
| 583 | hit |= grep_tree(opt, pathspec, &tree, &name, 0, 0); |
| 584 | strbuf_setlen(&name, name_base_len); |
| 585 | strbuf_addstr(&name, ce->name); |
| 586 | free(data); |
| 587 | } else if (S_ISREG(ce->ce_mode) && |
| 588 | match_pathspec(repo->index, pathspec, name.buf, name.len, 0, NULL, |
| 589 | S_ISDIR(ce->ce_mode) || |
| 590 | S_ISGITLINK(ce->ce_mode))) { |
| 591 | /* |
| 592 | * If CE_VALID is on, we assume worktree file and its |
| 593 | * cache entry are identical, even if worktree file has |
| 594 | * been modified, so use cache version instead |
| 595 | */ |
| 596 | if (cached || (ce->ce_flags & CE_VALID)) { |
| 597 | if (ce_stage(ce) || ce_intent_to_add(ce)) |
| 598 | continue; |
| 599 | hit |= grep_oid(opt, &ce->oid, name.buf, |
| 600 | 0, name.buf); |
| 601 | } else { |
| 602 | hit |= grep_file(opt, name.buf); |
| 603 | } |
| 604 | } else if (recurse_submodules && S_ISGITLINK(ce->ce_mode) && |
| 605 | submodule_path_match(repo->index, pathspec, name.buf, NULL)) { |
| 606 | hit |= grep_submodule(opt, pathspec, NULL, ce->name, |
| 607 | ce->name, cached); |
| 608 | } else { |
| 609 | continue; |
| 610 | } |
| 611 | |
| 612 | if (ce_stage(ce)) { |
| 613 | do { |
| 614 | nr++; |
| 615 | } while (nr < repo->index->cache_nr && |
| 616 | !strcmp(ce->name, repo->index->cache[nr]->name)); |
| 617 | nr--; /* compensate for loop control */ |
| 618 | } |
| 619 | if (hit && opt->status_only) |
| 620 | break; |
| 621 | } |
| 622 | |
| 623 | strbuf_release(&name); |
| 624 | return hit; |
| 625 | } |
| 626 | |
| 627 | static int grep_tree(struct grep_opt *opt, const struct pathspec *pathspec, |
| 628 | struct tree_desc *tree, struct strbuf *base, int tn_len, |
| 629 | int check_attr) |
| 630 | { |
| 631 | struct repository *repo = opt->repo; |
| 632 | int hit = 0; |
| 633 | enum interesting match = entry_not_interesting; |
| 634 | struct name_entry entry; |
| 635 | int old_baselen = base->len; |
| 636 | struct strbuf name = STRBUF_INIT; |
| 637 | int name_base_len = 0; |
| 638 | if (repo->submodule_prefix) { |
| 639 | strbuf_addstr(&name, repo->submodule_prefix); |
| 640 | name_base_len = name.len; |
| 641 | } |
| 642 | |
| 643 | while (tree_entry(tree, &entry)) { |
| 644 | int te_len = tree_entry_len(&entry); |
| 645 | |
| 646 | if (match != all_entries_interesting) { |
| 647 | strbuf_addstr(&name, base->buf + tn_len); |
| 648 | match = tree_entry_interesting(repo->index, |
| 649 | &entry, &name, |
| 650 | pathspec); |
| 651 | strbuf_setlen(&name, name_base_len); |
| 652 | |
| 653 | if (match == all_entries_not_interesting) |
| 654 | break; |
| 655 | if (match == entry_not_interesting) |
| 656 | continue; |
| 657 | } |
| 658 | |
| 659 | strbuf_add(base, entry.path, te_len); |
| 660 | |
| 661 | if (S_ISREG(entry.mode)) { |
| 662 | hit |= grep_oid(opt, &entry.oid, base->buf, tn_len, |
| 663 | check_attr ? base->buf + tn_len : NULL); |
| 664 | } else if (S_ISDIR(entry.mode)) { |
| 665 | enum object_type type; |
| 666 | struct tree_desc sub; |
| 667 | void *data; |
| 668 | size_t size; |
| 669 | |
| 670 | data = odb_read_object(the_repository->objects, |
| 671 | &entry.oid, &type, &size); |
| 672 | if (!data) |
| 673 | die(_("unable to read tree (%s)"), |
| 674 | oid_to_hex(&entry.oid)); |
| 675 | |
| 676 | strbuf_addch(base, '/'); |
| 677 | init_tree_desc(&sub, &entry.oid, data, size); |
| 678 | hit |= grep_tree(opt, pathspec, &sub, base, tn_len, |
| 679 | check_attr); |
| 680 | free(data); |
| 681 | } else if (recurse_submodules && S_ISGITLINK(entry.mode)) { |
| 682 | hit |= grep_submodule(opt, pathspec, &entry.oid, |
| 683 | base->buf, base->buf + tn_len, |
| 684 | 1); /* ignored */ |
| 685 | } |
| 686 | |
| 687 | strbuf_setlen(base, old_baselen); |
| 688 | |
| 689 | if (hit && opt->status_only) |
| 690 | break; |
| 691 | } |
| 692 | |
| 693 | strbuf_release(&name); |
| 694 | return hit; |
| 695 | } |
| 696 | |
| 697 | static void collect_blob_oids_for_tree(struct repository *repo, |
| 698 | const struct pathspec *pathspec, |
| 699 | struct tree_desc *tree, |
| 700 | struct strbuf *base, |
| 701 | int tn_len, |
| 702 | struct oidset *blob_oids) |
| 703 | { |
| 704 | struct name_entry entry; |
| 705 | int old_baselen = base->len; |
| 706 | struct strbuf name = STRBUF_INIT; |
| 707 | enum interesting match = entry_not_interesting; |
| 708 | |
| 709 | while (tree_entry(tree, &entry)) { |
| 710 | if (match != all_entries_interesting) { |
| 711 | strbuf_addstr(&name, base->buf + tn_len); |
| 712 | match = tree_entry_interesting(repo->index, |
| 713 | &entry, &name, |
| 714 | pathspec); |
| 715 | strbuf_reset(&name); |
| 716 | |
| 717 | if (match == all_entries_not_interesting) |
| 718 | break; |
| 719 | if (match == entry_not_interesting) |
| 720 | continue; |
| 721 | } |
| 722 | |
| 723 | strbuf_add(base, entry.path, tree_entry_len(&entry)); |
| 724 | |
| 725 | if (S_ISREG(entry.mode)) { |
| 726 | if (!odb_has_object(repo->objects, &entry.oid, 0)) |
| 727 | oidset_insert(blob_oids, &entry.oid); |
| 728 | } else if (S_ISDIR(entry.mode)) { |
| 729 | enum object_type type; |
| 730 | struct tree_desc sub_tree; |
| 731 | void *data; |
| 732 | size_t size; |
| 733 | |
| 734 | data = odb_read_object(repo->objects, &entry.oid, |
| 735 | &type, &size); |
| 736 | if (!data) |
| 737 | die(_("unable to read tree (%s)"), |
| 738 | oid_to_hex(&entry.oid)); |
| 739 | |
| 740 | strbuf_addch(base, '/'); |
| 741 | init_tree_desc(&sub_tree, &entry.oid, data, size); |
| 742 | collect_blob_oids_for_tree(repo, pathspec, &sub_tree, |
| 743 | base, tn_len, blob_oids); |
| 744 | free(data); |
| 745 | } |
| 746 | /* |
| 747 | * ...no else clause for S_ISGITLINK: submodules have their |
| 748 | * own promisor configuration and would need separate fetches |
| 749 | * anyway. |
| 750 | */ |
| 751 | |
| 752 | strbuf_setlen(base, old_baselen); |
| 753 | } |
| 754 | |
| 755 | strbuf_release(&name); |
| 756 | } |
| 757 | |
| 758 | static void collect_blob_oids_for_treeish(struct grep_opt *opt, |
| 759 | const struct pathspec *pathspec, |
| 760 | const struct object_id *tree_ish_oid, |
| 761 | const char *name, |
| 762 | struct oidset *blob_oids) |
| 763 | { |
| 764 | struct tree_desc tree; |
| 765 | void *data; |
| 766 | size_t size; |
| 767 | struct strbuf base = STRBUF_INIT; |
| 768 | int len; |
| 769 | |
| 770 | data = odb_read_object_peeled(opt->repo->objects, tree_ish_oid, |
| 771 | OBJ_TREE, &size, NULL); |
| 772 | |
| 773 | if (!data) |
| 774 | return; |
| 775 | |
| 776 | len = name ? strlen(name) : 0; |
| 777 | if (len) { |
| 778 | strbuf_add(&base, name, len); |
| 779 | strbuf_addch(&base, ':'); |
| 780 | } |
| 781 | init_tree_desc(&tree, tree_ish_oid, data, size); |
| 782 | |
| 783 | collect_blob_oids_for_tree(opt->repo, pathspec, &tree, |
| 784 | &base, base.len, blob_oids); |
| 785 | |
| 786 | strbuf_release(&base); |
| 787 | free(data); |
| 788 | } |
| 789 | |
| 790 | static void prefetch_grep_blobs(struct grep_opt *opt, |
| 791 | const struct pathspec *pathspec, |
| 792 | const struct object_array *list) |
| 793 | { |
| 794 | struct oidset blob_oids = OIDSET_INIT; |
| 795 | |
| 796 | /* Exit if we're not in a partial clone */ |
| 797 | if (!repo_has_promisor_remote(opt->repo)) |
| 798 | return; |
| 799 | |
| 800 | /* For each tree, gather the blobs in it */ |
| 801 | for (int i = 0; i < list->nr; i++) { |
| 802 | struct object *real_obj; |
| 803 | |
| 804 | obj_read_lock(); |
| 805 | real_obj = deref_tag(opt->repo, list->objects[i].item, |
| 806 | NULL, 0); |
| 807 | obj_read_unlock(); |
| 808 | |
| 809 | if (real_obj && |
| 810 | (real_obj->type == OBJ_COMMIT || |
| 811 | real_obj->type == OBJ_TREE)) |
| 812 | collect_blob_oids_for_treeish(opt, pathspec, |
| 813 | &real_obj->oid, |
| 814 | list->objects[i].name, |
| 815 | &blob_oids); |
| 816 | } |
| 817 | |
| 818 | /* Prefetch the blobs we found */ |
| 819 | if (oidset_size(&blob_oids)) { |
| 820 | struct oid_array to_fetch = OID_ARRAY_INIT; |
| 821 | struct oidset_iter iter; |
| 822 | const struct object_id *oid; |
| 823 | |
| 824 | oidset_iter_init(&blob_oids, &iter); |
| 825 | while ((oid = oidset_iter_next(&iter))) |
| 826 | oid_array_append(&to_fetch, oid); |
| 827 | |
| 828 | promisor_remote_get_direct(opt->repo, to_fetch.oid, to_fetch.nr); |
| 829 | |
| 830 | oid_array_clear(&to_fetch); |
| 831 | } |
| 832 | oidset_clear(&blob_oids); |
| 833 | } |
| 834 | |
| 835 | static int grep_object(struct grep_opt *opt, const struct pathspec *pathspec, |
| 836 | struct object *obj, const char *name, const char *path) |
| 837 | { |
| 838 | if (obj->type == OBJ_BLOB) |
| 839 | return grep_oid(opt, &obj->oid, name, 0, path); |
| 840 | if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) { |
| 841 | struct tree_desc tree; |
| 842 | void *data; |
| 843 | size_t size; |
| 844 | struct strbuf base; |
| 845 | int hit, len; |
| 846 | |
| 847 | data = odb_read_object_peeled(opt->repo->objects, &obj->oid, |
| 848 | OBJ_TREE, &size, NULL); |
| 849 | if (!data) |
| 850 | die(_("unable to read tree (%s)"), oid_to_hex(&obj->oid)); |
| 851 | |
| 852 | len = name ? strlen(name) : 0; |
| 853 | strbuf_init(&base, PATH_MAX + len + 1); |
| 854 | if (len) { |
| 855 | strbuf_add(&base, name, len); |
| 856 | strbuf_addch(&base, ':'); |
| 857 | } |
| 858 | init_tree_desc(&tree, &obj->oid, data, size); |
| 859 | hit = grep_tree(opt, pathspec, &tree, &base, base.len, |
| 860 | obj->type == OBJ_COMMIT); |
| 861 | strbuf_release(&base); |
| 862 | free(data); |
| 863 | return hit; |
| 864 | } |
| 865 | die(_("unable to grep from object of type %s"), type_name(obj->type)); |
| 866 | } |
| 867 | |
| 868 | static int grep_objects(struct grep_opt *opt, const struct pathspec *pathspec, |
| 869 | const struct object_array *list) |
| 870 | { |
| 871 | unsigned int i; |
| 872 | int hit = 0; |
| 873 | const unsigned int nr = list->nr; |
| 874 | |
| 875 | prefetch_grep_blobs(opt, pathspec, list); |
| 876 | |
| 877 | for (i = 0; i < nr; i++) { |
| 878 | struct object *real_obj; |
| 879 | |
| 880 | obj_read_lock(); |
| 881 | real_obj = deref_tag(opt->repo, list->objects[i].item, |
| 882 | NULL, 0); |
| 883 | obj_read_unlock(); |
| 884 | |
| 885 | if (!real_obj) { |
| 886 | char hex[GIT_MAX_HEXSZ + 1]; |
| 887 | const char *name = list->objects[i].name; |
| 888 | |
| 889 | if (!name) { |
| 890 | oid_to_hex_r(hex, &list->objects[i].item->oid); |
| 891 | name = hex; |
| 892 | } |
| 893 | die(_("invalid object '%s' given."), name); |
| 894 | } |
| 895 | |
| 896 | /* load the gitmodules file for this rev */ |
| 897 | if (recurse_submodules) { |
| 898 | submodule_free(opt->repo); |
| 899 | obj_read_lock(); |
| 900 | gitmodules_config_oid(&real_obj->oid); |
| 901 | obj_read_unlock(); |
| 902 | } |
| 903 | if (grep_object(opt, pathspec, real_obj, list->objects[i].name, |
| 904 | list->objects[i].path)) { |
| 905 | hit = 1; |
| 906 | if (opt->status_only) |
| 907 | break; |
| 908 | } |
| 909 | } |
| 910 | return hit; |
| 911 | } |
| 912 | |
| 913 | static int grep_directory(struct grep_opt *opt, const struct pathspec *pathspec, |
| 914 | int exc_std, int use_index) |
| 915 | { |
| 916 | struct dir_struct dir = DIR_INIT; |
| 917 | int i, hit = 0; |
| 918 | |
| 919 | if (!use_index) |
| 920 | dir.flags |= DIR_NO_GITLINKS; |
| 921 | if (exc_std) |
| 922 | setup_standard_excludes(&dir); |
| 923 | |
| 924 | fill_directory(&dir, opt->repo->index, pathspec); |
| 925 | for (i = 0; i < dir.nr; i++) { |
| 926 | hit |= grep_file(opt, dir.entries[i]->name); |
| 927 | if (hit && opt->status_only) |
| 928 | break; |
| 929 | } |
| 930 | dir_clear(&dir); |
| 931 | return hit; |
| 932 | } |
| 933 | |
| 934 | static int context_callback(const struct option *opt, const char *arg, |
| 935 | int unset) |
| 936 | { |
| 937 | struct grep_opt *grep_opt = opt->value; |
| 938 | int value; |
| 939 | const char *endp; |
| 940 | |
| 941 | if (unset) { |
| 942 | grep_opt->pre_context = grep_opt->post_context = 0; |
| 943 | return 0; |
| 944 | } |
| 945 | value = strtol(arg, (char **)&endp, 10); |
| 946 | if (*endp) { |
| 947 | return error(_("switch `%c' expects a numerical value"), |
| 948 | opt->short_name); |
| 949 | } |
| 950 | grep_opt->pre_context = grep_opt->post_context = value; |
| 951 | return 0; |
| 952 | } |
| 953 | |
| 954 | static int file_callback(const struct option *opt, const char *arg, int unset) |
| 955 | { |
| 956 | struct grep_opt *grep_opt = opt->value; |
| 957 | int from_stdin; |
| 958 | const char *filename = arg; |
| 959 | FILE *patterns; |
| 960 | int lno = 0; |
| 961 | struct strbuf sb = STRBUF_INIT; |
| 962 | |
| 963 | BUG_ON_OPT_NEG(unset); |
| 964 | |
| 965 | if (!*filename) |
| 966 | ; /* leave it as-is */ |
| 967 | else |
| 968 | filename = prefix_filename_except_for_dash(grep_prefix, filename); |
| 969 | |
| 970 | from_stdin = !strcmp(filename, "-"); |
| 971 | patterns = from_stdin ? stdin : fopen(filename, "r"); |
| 972 | if (!patterns) |
| 973 | die_errno(_("cannot open '%s'"), arg); |
| 974 | while (strbuf_getline(&sb, patterns) == 0) { |
| 975 | /* ignore empty line like grep does */ |
| 976 | if (sb.len == 0) |
| 977 | continue; |
| 978 | |
| 979 | append_grep_pat(grep_opt, sb.buf, sb.len, arg, ++lno, |
| 980 | GREP_PATTERN); |
| 981 | } |
| 982 | if (!from_stdin) |
| 983 | fclose(patterns); |
| 984 | strbuf_release(&sb); |
| 985 | if (filename != arg) |
| 986 | free((void *)filename); |
| 987 | return 0; |
| 988 | } |
| 989 | |
| 990 | static int not_callback(const struct option *opt, const char *arg, int unset) |
| 991 | { |
| 992 | struct grep_opt *grep_opt = opt->value; |
| 993 | BUG_ON_OPT_NEG(unset); |
| 994 | BUG_ON_OPT_ARG(arg); |
| 995 | append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT); |
| 996 | return 0; |
| 997 | } |
| 998 | |
| 999 | static int and_callback(const struct option *opt, const char *arg, int unset) |
| 1000 | { |
| 1001 | struct grep_opt *grep_opt = opt->value; |
| 1002 | BUG_ON_OPT_NEG(unset); |
| 1003 | BUG_ON_OPT_ARG(arg); |
| 1004 | append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND); |
| 1005 | return 0; |
| 1006 | } |
| 1007 | |
| 1008 | static int open_callback(const struct option *opt, const char *arg, int unset) |
| 1009 | { |
| 1010 | struct grep_opt *grep_opt = opt->value; |
| 1011 | BUG_ON_OPT_NEG(unset); |
| 1012 | BUG_ON_OPT_ARG(arg); |
| 1013 | append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN); |
| 1014 | return 0; |
| 1015 | } |
| 1016 | |
| 1017 | static int close_callback(const struct option *opt, const char *arg, int unset) |
| 1018 | { |
| 1019 | struct grep_opt *grep_opt = opt->value; |
| 1020 | BUG_ON_OPT_NEG(unset); |
| 1021 | BUG_ON_OPT_ARG(arg); |
| 1022 | append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN); |
| 1023 | return 0; |
| 1024 | } |
| 1025 | |
| 1026 | static int pattern_callback(const struct option *opt, const char *arg, |
| 1027 | int unset) |
| 1028 | { |
| 1029 | struct grep_opt *grep_opt = opt->value; |
| 1030 | BUG_ON_OPT_NEG(unset); |
| 1031 | append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN); |
| 1032 | return 0; |
| 1033 | } |
| 1034 | |
| 1035 | int cmd_grep(int argc, |
| 1036 | const char **argv, |
| 1037 | const char *prefix, |
| 1038 | struct repository *repo UNUSED) |
| 1039 | { |
| 1040 | int hit = 0; |
| 1041 | int cached = 0, untracked = 0, opt_exclude = -1; |
| 1042 | int seen_dashdash = 0; |
| 1043 | int external_grep_allowed__ignored; |
| 1044 | const char *show_in_pager = NULL, *default_pager = "dummy"; |
| 1045 | struct grep_opt opt; |
| 1046 | struct object_array list = OBJECT_ARRAY_INIT; |
| 1047 | struct pathspec pathspec; |
| 1048 | struct string_list path_list = STRING_LIST_INIT_DUP; |
| 1049 | int i; |
| 1050 | int dummy; |
| 1051 | int use_index = 1; |
| 1052 | int allow_revs; |
| 1053 | int ret; |
| 1054 | |
| 1055 | struct option options[] = { |
| 1056 | OPT_BOOL(0, "cached", &cached, |
| 1057 | N_("search in index instead of in the work tree")), |
| 1058 | OPT_NEGBIT(0, "no-index", &use_index, |
| 1059 | N_("find in contents not managed by git"), 1), |
| 1060 | OPT_BOOL(0, "untracked", &untracked, |
| 1061 | N_("search in both tracked and untracked files")), |
| 1062 | OPT_SET_INT(0, "exclude-standard", &opt_exclude, |
| 1063 | N_("ignore files specified via '.gitignore'"), 1), |
| 1064 | OPT_BOOL(0, "recurse-submodules", &recurse_submodules, |
| 1065 | N_("recursively search in each submodule")), |
| 1066 | OPT_GROUP(""), |
| 1067 | OPT_BOOL('v', "invert-match", &opt.invert, |
| 1068 | N_("show non-matching lines")), |
| 1069 | OPT_BOOL('i', "ignore-case", &opt.ignore_case, |
| 1070 | N_("case insensitive matching")), |
| 1071 | OPT_BOOL('w', "word-regexp", &opt.word_regexp, |
| 1072 | N_("match patterns only at word boundaries")), |
| 1073 | OPT_SET_INT('a', "text", &opt.binary, |
| 1074 | N_("process binary files as text"), GREP_BINARY_TEXT), |
| 1075 | OPT_SET_INT('I', NULL, &opt.binary, |
| 1076 | N_("don't match patterns in binary files"), |
| 1077 | GREP_BINARY_NOMATCH), |
| 1078 | OPT_BOOL(0, "textconv", &opt.allow_textconv, |
| 1079 | N_("process binary files with textconv filters")), |
| 1080 | OPT_SET_INT('r', "recursive", &opt.max_depth, |
| 1081 | N_("search in subdirectories (default)"), -1), |
| 1082 | OPT_INTEGER_F(0, "max-depth", &opt.max_depth, |
| 1083 | N_("descend at most <n> levels"), PARSE_OPT_NONEG), |
| 1084 | OPT_GROUP(""), |
| 1085 | OPT_SET_INT('E', "extended-regexp", &opt.pattern_type_option, |
| 1086 | N_("use extended POSIX regular expressions"), |
| 1087 | GREP_PATTERN_TYPE_ERE), |
| 1088 | OPT_SET_INT('G', "basic-regexp", &opt.pattern_type_option, |
| 1089 | N_("use basic POSIX regular expressions (default)"), |
| 1090 | GREP_PATTERN_TYPE_BRE), |
| 1091 | OPT_SET_INT('F', "fixed-strings", &opt.pattern_type_option, |
| 1092 | N_("interpret patterns as fixed strings"), |
| 1093 | GREP_PATTERN_TYPE_FIXED), |
| 1094 | OPT_SET_INT('P', "perl-regexp", &opt.pattern_type_option, |
| 1095 | N_("use Perl-compatible regular expressions"), |
| 1096 | GREP_PATTERN_TYPE_PCRE), |
| 1097 | OPT_GROUP(""), |
| 1098 | OPT_BOOL('n', "line-number", &opt.linenum, N_("show line numbers")), |
| 1099 | OPT_BOOL(0, "column", &opt.columnnum, N_("show column number of first match")), |
| 1100 | OPT_NEGBIT('h', NULL, &opt.pathname, N_("don't show filenames"), 1), |
| 1101 | OPT_BIT('H', NULL, &opt.pathname, N_("show filenames"), 1), |
| 1102 | OPT_NEGBIT(0, "full-name", &opt.relative, |
| 1103 | N_("show filenames relative to top directory"), 1), |
| 1104 | OPT_BOOL('l', "files-with-matches", &opt.name_only, |
| 1105 | N_("show only filenames instead of matching lines")), |
| 1106 | OPT_BOOL(0, "name-only", &opt.name_only, |
| 1107 | N_("synonym for --files-with-matches")), |
| 1108 | OPT_BOOL('L', "files-without-match", |
| 1109 | &opt.unmatch_name_only, |
| 1110 | N_("show only the names of files without match")), |
| 1111 | OPT_BOOL_F('z', "null", &opt.null_following_name, |
| 1112 | N_("print NUL after filenames"), |
| 1113 | PARSE_OPT_NOCOMPLETE), |
| 1114 | OPT_BOOL('o', "only-matching", &opt.only_matching, |
| 1115 | N_("show only matching parts of a line")), |
| 1116 | OPT_BOOL('c', "count", &opt.count, |
| 1117 | N_("show the number of matches instead of matching lines")), |
| 1118 | OPT__COLOR(&opt.color, N_("highlight matches")), |
| 1119 | OPT_BOOL(0, "break", &opt.file_break, |
| 1120 | N_("print empty line between matches from different files")), |
| 1121 | OPT_BOOL(0, "heading", &opt.heading, |
| 1122 | N_("show filename only once above matches from same file")), |
| 1123 | OPT_GROUP(""), |
| 1124 | OPT_CALLBACK('C', "context", &opt, N_("n"), |
| 1125 | N_("show <n> context lines before and after matches"), |
| 1126 | context_callback), |
| 1127 | OPT_UNSIGNED('B', "before-context", &opt.pre_context, |
| 1128 | N_("show <n> context lines before matches")), |
| 1129 | OPT_UNSIGNED('A', "after-context", &opt.post_context, |
| 1130 | N_("show <n> context lines after matches")), |
| 1131 | OPT_INTEGER(0, "threads", &num_threads, |
| 1132 | N_("use <n> worker threads")), |
| 1133 | OPT_NUMBER_CALLBACK(&opt, N_("shortcut for -C NUM"), |
| 1134 | context_callback), |
| 1135 | OPT_BOOL('p', "show-function", &opt.funcname, |
| 1136 | N_("show a line with the function name before matches")), |
| 1137 | OPT_BOOL('W', "function-context", &opt.funcbody, |
| 1138 | N_("show the surrounding function")), |
| 1139 | OPT_GROUP(""), |
| 1140 | OPT_CALLBACK('f', NULL, &opt, N_("file"), |
| 1141 | N_("read patterns from file"), file_callback), |
| 1142 | OPT_CALLBACK_F('e', NULL, &opt, N_("pattern"), |
| 1143 | N_("match <pattern>"), PARSE_OPT_NONEG, pattern_callback), |
| 1144 | OPT_CALLBACK_F(0, "and", &opt, NULL, |
| 1145 | N_("combine patterns specified with -e"), |
| 1146 | PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback), |
| 1147 | OPT_BOOL_F(0, "or", &dummy, "", PARSE_OPT_NONEG), |
| 1148 | OPT_CALLBACK_F(0, "not", &opt, NULL, "", |
| 1149 | PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback), |
| 1150 | OPT_CALLBACK_F('(', NULL, &opt, NULL, "", |
| 1151 | PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH, |
| 1152 | open_callback), |
| 1153 | OPT_CALLBACK_F(')', NULL, &opt, NULL, "", |
| 1154 | PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH, |
| 1155 | close_callback), |
| 1156 | OPT__QUIET(&opt.status_only, |
| 1157 | N_("indicate hit with exit status without output")), |
| 1158 | OPT_BOOL(0, "all-match", &opt.all_match, |
| 1159 | N_("show only matches from files that match all patterns")), |
| 1160 | OPT_GROUP(""), |
| 1161 | { |
| 1162 | .type = OPTION_STRING, |
| 1163 | .short_name = 'O', |
| 1164 | .long_name = "open-files-in-pager", |
| 1165 | .value = &show_in_pager, |
| 1166 | .argh = N_("pager"), |
| 1167 | .help = N_("show matching files in the pager"), |
| 1168 | .flags = PARSE_OPT_OPTARG | PARSE_OPT_NOCOMPLETE, |
| 1169 | .defval = (intptr_t)default_pager, |
| 1170 | }, |
| 1171 | OPT_BOOL_F(0, "ext-grep", &external_grep_allowed__ignored, |
| 1172 | N_("allow calling of grep(1) (ignored by this build)"), |
| 1173 | PARSE_OPT_NOCOMPLETE), |
| 1174 | OPT_INTEGER('m', "max-count", &opt.max_count, |
| 1175 | N_("maximum number of results per file")), |
| 1176 | OPT_END() |
| 1177 | }; |
| 1178 | grep_prefix = prefix; |
| 1179 | |
| 1180 | grep_init(&opt, the_repository); |
| 1181 | repo_config(the_repository, grep_cmd_config, &opt); |
| 1182 | |
| 1183 | /* |
| 1184 | * If there is no -- then the paths must exist in the working |
| 1185 | * tree. If there is no explicit pattern specified with -e or |
| 1186 | * -f, we take the first unrecognized non option to be the |
| 1187 | * pattern, but then what follows it must be zero or more |
| 1188 | * valid refs up to the -- (if exists), and then existing |
| 1189 | * paths. If there is an explicit pattern, then the first |
| 1190 | * unrecognized non option is the beginning of the refs list |
| 1191 | * that continues up to the -- (if exists), and then paths. |
| 1192 | */ |
| 1193 | argc = parse_options(argc, argv, prefix, options, grep_usage, |
| 1194 | PARSE_OPT_KEEP_DASHDASH | |
| 1195 | PARSE_OPT_STOP_AT_NON_OPTION); |
| 1196 | |
| 1197 | if (the_repository->gitdir) { |
| 1198 | prepare_repo_settings(the_repository); |
| 1199 | the_repository->settings.command_requires_full_index = 0; |
| 1200 | } |
| 1201 | |
| 1202 | if (use_index && !startup_info->have_repository) { |
| 1203 | int fallback = 0; |
| 1204 | repo_config_get_bool(the_repository, "grep.fallbacktonoindex", &fallback); |
| 1205 | if (fallback) |
| 1206 | use_index = 0; |
| 1207 | else |
| 1208 | /* die the same way as if we did it at the beginning */ |
| 1209 | setup_git_directory(the_repository); |
| 1210 | } |
| 1211 | /* Ignore --recurse-submodules if --no-index is given or implied */ |
| 1212 | if (!use_index) |
| 1213 | recurse_submodules = 0; |
| 1214 | |
| 1215 | /* |
| 1216 | * skip a -- separator; we know it cannot be |
| 1217 | * separating revisions from pathnames if |
| 1218 | * we haven't even had any patterns yet |
| 1219 | */ |
| 1220 | if (argc > 0 && !opt.pattern_list && !strcmp(argv[0], "--")) { |
| 1221 | argv++; |
| 1222 | argc--; |
| 1223 | } |
| 1224 | |
| 1225 | /* First unrecognized non-option token */ |
| 1226 | if (argc > 0 && !opt.pattern_list) { |
| 1227 | append_grep_pattern(&opt, argv[0], "command line", 0, |
| 1228 | GREP_PATTERN); |
| 1229 | argv++; |
| 1230 | argc--; |
| 1231 | } |
| 1232 | |
| 1233 | if (show_in_pager == default_pager) |
| 1234 | show_in_pager = git_pager(the_repository, 1); |
| 1235 | if (show_in_pager) { |
| 1236 | opt.color = GIT_COLOR_NEVER; |
| 1237 | opt.name_only = 1; |
| 1238 | opt.null_following_name = 1; |
| 1239 | opt.output_priv = &path_list; |
| 1240 | opt.output = append_path; |
| 1241 | string_list_append(&path_list, show_in_pager); |
| 1242 | } |
| 1243 | |
| 1244 | if (!opt.pattern_list) |
| 1245 | die(_("no pattern given")); |
| 1246 | |
| 1247 | /* --only-matching has no effect with --invert. */ |
| 1248 | if (opt.invert) |
| 1249 | opt.only_matching = 0; |
| 1250 | |
| 1251 | /* |
| 1252 | * We have to find "--" in a separate pass, because its presence |
| 1253 | * influences how we will parse arguments that come before it. |
| 1254 | */ |
| 1255 | for (i = 0; i < argc; i++) { |
| 1256 | if (!strcmp(argv[i], "--")) { |
| 1257 | seen_dashdash = 1; |
| 1258 | break; |
| 1259 | } |
| 1260 | } |
| 1261 | |
| 1262 | /* |
| 1263 | * Resolve any rev arguments. If we have a dashdash, then everything up |
| 1264 | * to it must resolve as a rev. If not, then we stop at the first |
| 1265 | * non-rev and assume everything else is a path. |
| 1266 | */ |
| 1267 | allow_revs = use_index && !untracked; |
| 1268 | for (i = 0; i < argc; i++) { |
| 1269 | const char *arg = argv[i]; |
| 1270 | struct object_id oid; |
| 1271 | struct object_context oc = {0}; |
| 1272 | struct object *object; |
| 1273 | |
| 1274 | if (!strcmp(arg, "--")) { |
| 1275 | i++; |
| 1276 | break; |
| 1277 | } |
| 1278 | |
| 1279 | if (!allow_revs) { |
| 1280 | if (seen_dashdash) |
| 1281 | die(_("--no-index or --untracked cannot be used with revs")); |
| 1282 | break; |
| 1283 | } |
| 1284 | |
| 1285 | if (get_oid_with_context(the_repository, arg, |
| 1286 | GET_OID_RECORD_PATH, |
| 1287 | &oid, &oc)) { |
| 1288 | if (seen_dashdash) |
| 1289 | die(_("unable to resolve revision: %s"), arg); |
| 1290 | object_context_release(&oc); |
| 1291 | break; |
| 1292 | } |
| 1293 | |
| 1294 | object = parse_object_or_die(the_repository, &oid, arg); |
| 1295 | if (!seen_dashdash) |
| 1296 | verify_non_filename(the_repository, prefix, arg); |
| 1297 | add_object_array_with_path(object, arg, &list, oc.mode, oc.path); |
| 1298 | object_context_release(&oc); |
| 1299 | } |
| 1300 | |
| 1301 | /* |
| 1302 | * Anything left over is presumed to be a path. But in the non-dashdash |
| 1303 | * "do what I mean" case, we verify and complain when that isn't true. |
| 1304 | */ |
| 1305 | if (!seen_dashdash) { |
| 1306 | int j; |
| 1307 | for (j = i; j < argc; j++) |
| 1308 | verify_filename(the_repository, prefix, argv[j], j == i && allow_revs); |
| 1309 | } |
| 1310 | |
| 1311 | parse_pathspec(&pathspec, 0, |
| 1312 | PATHSPEC_PREFER_CWD | |
| 1313 | (opt.max_depth != -1 ? PATHSPEC_MAXDEPTH_VALID : 0), |
| 1314 | prefix, argv + i); |
| 1315 | pathspec.max_depth = opt.max_depth; |
| 1316 | pathspec.recursive = 1; |
| 1317 | pathspec.recurse_submodules = !!recurse_submodules; |
| 1318 | |
| 1319 | if (recurse_submodules && untracked) |
| 1320 | die(_("--untracked not supported with --recurse-submodules")); |
| 1321 | |
| 1322 | /* |
| 1323 | * Optimize out the case where the amount of matches is limited to zero. |
| 1324 | * We do this to keep results consistent with GNU grep(1). |
| 1325 | */ |
| 1326 | if (opt.max_count == 0) { |
| 1327 | ret = 1; |
| 1328 | goto out; |
| 1329 | } |
| 1330 | |
| 1331 | if (show_in_pager) { |
| 1332 | if (num_threads > 1) |
| 1333 | warning(_("invalid option combination, ignoring --threads")); |
| 1334 | num_threads = 1; |
| 1335 | } else if (!HAVE_THREADS && num_threads > 1) { |
| 1336 | warning(_("no threads support, ignoring --threads")); |
| 1337 | num_threads = 1; |
| 1338 | } else if (num_threads < 0) |
| 1339 | die(_("invalid number of threads specified (%d)"), num_threads); |
| 1340 | else if (num_threads == 0) |
| 1341 | num_threads = HAVE_THREADS ? online_cpus() : 1; |
| 1342 | |
| 1343 | if (num_threads > 1) { |
| 1344 | if (!HAVE_THREADS) |
| 1345 | BUG("Somebody got num_threads calculation wrong!"); |
| 1346 | if (!(opt.name_only || opt.unmatch_name_only || opt.count) |
| 1347 | && (opt.pre_context || opt.post_context || |
| 1348 | opt.file_break || opt.funcbody)) |
| 1349 | skip_first_line = 1; |
| 1350 | |
| 1351 | /* |
| 1352 | * Pre-read gitmodules (if not read already) and force eager |
| 1353 | * initialization of packed_git to prevent racy lazy |
| 1354 | * reading/initialization once worker threads are started. |
| 1355 | */ |
| 1356 | if (recurse_submodules) |
| 1357 | repo_read_gitmodules(the_repository, 1); |
| 1358 | |
| 1359 | if (startup_info->have_repository) |
| 1360 | odb_prepare(the_repository->objects, 0); |
| 1361 | |
| 1362 | start_threads(&opt); |
| 1363 | } else { |
| 1364 | /* |
| 1365 | * The compiled patterns on the main path are only |
| 1366 | * used when not using threading. Otherwise |
| 1367 | * start_threads() above calls compile_grep_patterns() |
| 1368 | * for each thread. |
| 1369 | */ |
| 1370 | compile_grep_patterns(&opt); |
| 1371 | } |
| 1372 | |
| 1373 | if (show_in_pager && (cached || list.nr)) |
| 1374 | die(_("--open-files-in-pager only works on the worktree")); |
| 1375 | |
| 1376 | if (show_in_pager && opt.pattern_list && !opt.pattern_list->next) { |
| 1377 | const char *pager = path_list.items[0].string; |
| 1378 | int len = strlen(pager); |
| 1379 | |
| 1380 | if (len > 4 && is_dir_sep(pager[len - 5])) |
| 1381 | pager += len - 4; |
| 1382 | |
| 1383 | if (opt.ignore_case && !strcmp("less", pager)) |
| 1384 | string_list_append(&path_list, "-I"); |
| 1385 | |
| 1386 | if (!strcmp("less", pager) || !strcmp("vi", pager)) { |
| 1387 | struct strbuf buf = STRBUF_INIT; |
| 1388 | strbuf_addf(&buf, "+/%s%s", |
| 1389 | strcmp("less", pager) ? "" : "*", |
| 1390 | opt.pattern_list->pattern); |
| 1391 | string_list_append_nodup(&path_list, |
| 1392 | strbuf_detach(&buf, NULL)); |
| 1393 | } |
| 1394 | } |
| 1395 | |
| 1396 | if (!show_in_pager && !opt.status_only) |
| 1397 | setup_pager(the_repository); |
| 1398 | |
| 1399 | die_for_incompatible_opt3(!use_index, "--no-index", |
| 1400 | untracked, "--untracked", |
| 1401 | cached, "--cached"); |
| 1402 | |
| 1403 | if (!use_index || untracked) { |
| 1404 | int use_exclude = (opt_exclude < 0) ? use_index : !!opt_exclude; |
| 1405 | hit = grep_directory(&opt, &pathspec, use_exclude, use_index); |
| 1406 | } else if (0 <= opt_exclude) { |
| 1407 | die(_("--[no-]exclude-standard cannot be used for tracked contents")); |
| 1408 | } else if (!list.nr) { |
| 1409 | if (!cached) |
| 1410 | setup_work_tree(the_repository); |
| 1411 | |
| 1412 | hit = grep_cache(&opt, &pathspec, cached); |
| 1413 | } else { |
| 1414 | if (cached) |
| 1415 | die(_("both --cached and trees are given")); |
| 1416 | |
| 1417 | hit = grep_objects(&opt, &pathspec, &list); |
| 1418 | } |
| 1419 | |
| 1420 | if (num_threads > 1) |
| 1421 | hit |= wait_all(); |
| 1422 | if (hit && show_in_pager) |
| 1423 | run_pager(&opt, prefix); |
| 1424 | |
| 1425 | ret = !hit; |
| 1426 | |
| 1427 | out: |
| 1428 | clear_pathspec(&pathspec); |
| 1429 | string_list_clear(&path_list, 0); |
| 1430 | free_grep_patterns(&opt); |
| 1431 | object_array_clear(&list); |
| 1432 | free_repos(); |
| 1433 | return ret; |
| 1434 | } |